Surrogate endpoints are widely used in clinical trials, especially in situations where the endpoint of interest is not directly observable or to avoid long trial periods. A typical example for this case is frequently found in clinical trials in oncology, where overall survival (OS) as endpoint of interest and progression free survival (PFS) as surrogate endpoint are discriminated.
Based on the perspective of case definitions on surrogate endpoints, we provide a formal definition of such endpoints followed by a description of the structure of surrogate endpoints.
Surrogate endpoints can be considered as case definitions for the endpoint of interest. Therefore, the performance of surrogate endpoints can be described using the classical terminology of diagnostic tests including sensitivity and specificity. Since such endpoints always focus on sensitivity with necessarily reduced specificity, efficacy estimates based on such endpoints are in general biased.